Objective: To assess whether measures of body fat by DXA scanning can improve prediction of insulin sensitivity (S(I)) beyond what is possible with traditional measures, such as BMI, waist circumference, and waist-to-hip ratio (WHR).
Research methods and procedures: Frequently sampled intravenous glucose tolerance tests were performed in 256 asymptomatic non-Hispanic white subjects from Rochester, MN (age 19-60 years; 123 men and 133 women) to determine the S(I) index by Bergman's minimal model technique. Height, weight, and waist and hip circumferences were measured for calculation of BMI and WHR; DXA was used to determine fat in the head, upper body, abdomen, and lower body. Linear regression was used to assess their relationships with S(I) after sex stratification and adjustment for age.
Results: After controlling for age, increases in traditional and DXA measures of fat were consistently associated with smaller declines in S(I) among women than among men. In men, after controlling for age, all of the predictive information of S(I) was provided by waist circumference (additional R2 = 0.39, p < 0.001); none of the DXA measures improved the ability to predict S(I). In women, after adjustment for age, BMI, and WHR, the only DXA measure that improved the prediction of S(I) was percentage head fat (additional R2 = 0.03, p < 0.001).
Discussion: Equivalent increases in most measures of body fat had lesser impact on S(I) in women than in men. In both sexes, the predictive information provided by DXA measures is approximately equal to, but not additive to, that provided by simpler, traditional measures.